IEA Task 25 – Design and Operation of Power Systems with large

advertisement
IEA Task 25 – Design and
Operation of Power Systems
with large amounts of Wind
Power.
Elforsk rapport 09:07
Lennart Söder
April 2009
IEA Task 25 – Design and
Operation of Power Systems
with large amounts of Wind
Power
Swedish participation
Elforsk rapport 09:07
Lennart Söder
April 2009
ELFORSK
ELFORSK
Preface
Knowledge transfer from many countries with experience and research in the
area of integration of wind power into the power system is important.
Work with the IEA Task XXV "Design and operation of Power Systems with
large amounts of Wind Power" has therefore been carried out as a project
with the Vindforsk II program.
This report is a short presentation of the task's activities and contains a list of
published reports and papers within the project. It also contains a summary
from the final report of the IEA Task yet to be published. This report is
expected to be published in 2009.
The research project presented in this report was carried out by Lennart
Söder, KTH
The Swedish wind energy research programme "Vindforsk - II", which was
funded by ABB, the Norwegian based EBL-Kompetanse, E.ON Sverige AB,
Falkenberg Energi AB, Göteborg Energi, Jämtkraft AB, Karlstad Energi AB,
Luleå Energi AB, Lunds Energi AB, Skellefteå Kraft AB, Svenska Kraftnät,
Swedish Energy Agency, Tekniska Verken i Linköping AB, Umeå Energi AB,
Vattenfall AB and Öresundskraft AB and Energimyndigheten.
Stockholm April 2009
Anders Björck
Program leader of Vindforsk II
ELFORSK
ELFORSK
Sammanfattning
Ett internationellt forskningssamarbete kallat “Design and Operation of Power
Systems with Large Amounts of Wind Power” (Design och drift av kraftsystem
med stora mängder vindkraft) startades 2006 inom “IEA Implementing
Agreement on the Co-operation in the Research, Development and
Deployment of Wind Turbine Systems” (www.ieawind.org) såsom “Task 25”.
Detta forskningssamarbete samlar in och sprider information och erfarenheter
från studier rörande vindkraftens inverkan på kraftsystemet, samt granska
olika metoder, beräkningsverktyg och använda data. Resultaten från den
första 3-årsperioden presenteras i denna rapport. Arbetet kommer att
fortsätta med en andra 3-års-period. Följande länder och forskningsinstitut
har deltagit i samarbetet. (TSO = Transmission System Operator, dvs
driftansvarig organisation för transmissionsnätet):
•
Danmark: Risø-DTU; TSO Energinet.dk
•
EWEA (European Wind Energy Association)
•
Finland: VTT Technical Research Centre of Finland (Operating Agent)
•
Tyskland: ISET; Transmissionssystemoperatörerna RWE and E.ON Netz
•
Irland: SEI; UCD; ECAR; TSO Eirgrid
•
Norge: SINTEF; Statkraft
•
Nederländerna: ECN; TUDelft
•
Portugal: INETI; TSO REN; INESC-Porto, IST
•
Spanien: University Castilla La Mancha
•
Sverige: KTH
•
Storbritannien:
Centre
for
Distributed
Generation
&
Sustainable
Electrical Energy
•
USA: NREL; UWIG.
Forskningssamarbetet har producerat ett antal state-of-the-art rapporter vilka
har publicerats som ”VTT Working Papers series” sedan 2007. I dessa
rapporter, liksom i denna rapport presenteras endast slutsatser från nyligen
genomförda, avslutade studie. Samarbetet har även startat arbetet med att
utveckla rekommendationer för lämpliga metoder där syftet är att bedömma
systeminverkan och kostnader orsakade av integration av store mängder
vindkraft.
I denna rapport presenteras endast de överordnade slutsatserna från
slutrapporten från fas 1 i samarbetet. Hela slutrapporten (209 sidor) finns
tillgänglig på http://www.ieawind.org/AnnexXXV.html
ELFORSK
Vindforsk har finansierat den svenska medverkan i detta samarbete. Denna
medverkan har inneburit att personer från avdelningen för Elektriska
Energisystem, KTH, Stockholm har deltagit vid möten samt presenterat
forskningsresultat från KTH och andra studier rörande integration av vindkraft
i Sverige.
Följande är en översättning av ”Executive summary” från slutrapporten.
Utbyggnaden av vindkraft innebär att en varierande och endast delvis
prognoserbar energikälla ska integreras i kraftsystemet i vilket det hela tiden
måste upprätthållas en balans mellan produktion och konsumtion. Stora
mängder vindkraft har en inverkan på såväl anslutning, integration av
produktionen, planeringen av transmissionsnätet samt driften av elsystemet
och elmarknadens uppbyggnad. Denna slutrapport av Task 25, fas 1,
presenterar en sammanfattning av några utvalda, nyligen publicerad studier
av integration av vindkraft från de deltagande länderna. De olika studierna
jämförs, trots att detta inte är en trivial uppgift beroende på olika metoder,
olika data samt olika antaganden rörande, t ex, tillgänglig transmissionskapacitet.
Det finns redan idag flera kraftsystem där större mängder vindkraft
integrerats. Flera aspekter vilka påverkar mängden vindkraft som kan
integreras har identifierats. Sammanlagring innebär att den totala variationen
från vindkraft i olika geografiska områden minskar relativt sett på samma sätt
som att koordinering av balanseringsresurser i flera områden reducerar
kostnaden för balanseringen. Ett alternativ till att ha större områden för
balansering är att tillåta handel mellan olika areor för balansering inom
timmen eller inom dagen. Att tillåta drift och planering av elmarknaden på en
kortare tidshorisont än dygnet före hjälper till att minska behovet av
driftsreserver. Ett bra transmissionssystem är nyckeln till en effektiv
sammanlagring av vindkraften och till ett effektivt utnyttjande av tillgängliga
balanseringsresurser.
För
vindkraftnivåer
på
10-20%
av
total
energikonsumtion
är
kostnadseffektiviteten att använda ny lagringskapacitet relativt låg (förutom
vattenkraft med magasin eller pumpkraftverk). Vid större mängd vindkraft
kan den extra flexibilitet som lagring innebär vara rationell för driften av
elsystemet förutsatt att den är konkurrenskraftig gentemot annan typ av
flexibilitet. Det bör dock påpekas att all form av lager skall användas för hela
systemets behov. Det är inte kostnadseffektivt att erbjuda speciell back-up
just för vindkraften i större kraftsystem där den totala variationen från
elförbrukning och produktionsanläggningar sammanlagras, på samma sätt
som det inte är effektivt att ha speciella lager för haverier i enskilda
värmekraftverk, eller ha speciella kraftverk som följer variationen i en specifik
elförbrukning.
Integrationskostnad för vindkraft: Många studier tar upp frågan om
”integrationskostnad”. Denna kostnad är den extra kostnad för utbyggnad och
drift av icke-vind-delen av kraftsystemet som orsakas av utbyggnaden av
vindkraften. Integrationskostnaden kan delas upp i olika delar som härrör från
balanseringskostnaden samt förstärkning av elnätet. Det är dock viktigt att
beakta huruvida man uppskattar en ”marknadskostnad” eller en ”fysisk
ELFORSK
kostnad” för kraftsystemet. En ”marknadskostnad” kan innebära överföring av
medel från en aktör till en annan, medan en ”fysisk kostnad” innebär att
kostnaden för hela systemet ökar. De flesta studierna har hittills inriktats på
”kostnaden för integration”, medan det även finns ett behov av att också göra
kostnad-nytta-analyser. Det finns också en nytta med att integrera vindkraft:
den reducerar den totala driftskostnaden och dessutom reduceras utsläppen
om vindkraften ersätter fossila bränslen. Integrationskostnad för vindkraft
måste även jämföras med något, t ex produktionskostnaden eller
marknadsvärdet för vindkraft, eller alternativt integrationskostnaden för
andra produktionsslag. För att erhålla en rättvis jämförelse mellan
kraftsystem med olika mängder vindkraft, bör dessa system i princip ha
samma CO2-utsläpp, tillförlitlighet etc. Vindkraftens effektvärde måste
därmed även bli fastställt.
Ökning av korttidsreserver beroende på vindkraft: Vindkraft kan öka
kraven på systemoperatören att hålla mer driftsreserver. Både statistik och
studier visar att den huvudsakliga utmaningen är vindkraftens variation inom
1-6 timmar. Frekvensregleringen (tidsskala sekunder) och systemets elektromekaniska tröghet är inte den stora utmaningen för närvarande, men kan bli
en utmaning för små system med hög andel vindkraft i framtiden. Ökningen
av behovet av korttidsreserver uppskattas ofta med statistiska metoder där
man kombinerar variationen i prognosfel för elförbrukning och vindkraft för
att få fram den nettovariation som systemet ser. Inverkan av vindkraft kan
noteras från tidsskalan 10 minuter upp till några timmar, men endast lite i
tidsskalan sekund till sekund vilket är tidsskalan för den automatiska
frekvensregleringen.
Den
uppskattade
ökningen
av
behovet
av
korttidsreserver i de i denna rapport summerade studierna har en stor
spridning: 1-15% av installerad effekt vid en vindkraftsandel om 10% av
elförbrukningen samt 4-18% av installerad effekt vid en vindkraftsandel om
20%. Det är här centralt att separera mellan flexibiliteten vid längre tidsskalor
från flera timmar upp till ett dygn (kraftverk som följer elförbrukningen) och
behovet av reserver vilka kan aktiveras inom sekunder eller minuter
(kraftverk som följer oprognoserade variationer i nettoförbrukningen).
En viktig slutstats är att ”ökat behov av reserver” inte nödvändigtvis behöver
betyda behov av nya investeringar. Ett exempel är att behovet av vindorsakade reserver är störst när vindkraften producerar mycket. Vid dessa
situationer producerar de andra kraftverken vid en lägre nivå, vilket innebär
att de kan fungera som reserver och öka produktionen om vindkraften
minskar. Detta måste beaktas när ”ökat behov av reserver” ska uppskattas.
Från kostnadsuppskattningar som presenterats i studierna följer att vid en
vindkraftsandel om upp till 20% av elförbrukningen (årsenergi), kommer
systemkostnaden för vindkraftens balansering och osäkerhet att vara ungefär
1-4 Eurocent/MWh vindkraftsproduktion. Detta är 10% eller mindre av
försäljningsvärdet för vindkraften. Den faktiska inverkan av att integrera mer
vindkraft kan bero på flera olika lokala förutsättningar. Viktiga faktorer för att
reducera integrationskostnaden är sammanlagring av vindkraften över större
regioner, större geografiska områden för balansering, samt att planera driften
av kraftsystemet närmare inpå faktisk driftstimme då det finns tillgång till
bättre prognoser för faktisk vindkraftsproduktion.
ELFORSK
Transmissionsplanering med vindkraft: Med dagens teknik har nya
vindkraftverk möjlighet att möta nätägarnas krav/önskemål avseende att
fortsätta producera även efter ett kostvarigt spänningsfall (fault ride
through); att leverera reaktiv effekt till systemet; att styra spänningen; och
att via Scada-system styra ökning och minskning av effekten.
Nätförstärkningar kan behövas för att hantera större effektflöden och för att
bibehålla en stabil spänning. Detta kan vara nödvändigt när större mängder
vindkraft ansluts till svaga nät, när det finns flaskhalsar mellan produktion
och elförbrukning eller där det inte finns nät, som vid havsbaserad vindkraft.
Transmissionskostnad är här den extra kostnad i transmissionssystemet
somm uppstår när vindkraft integreras. Antingen hänförs all denna extra
kostnad till vindkraften eller också endast en viss del – nätförstärkningar och
nya ledningar gynnar ofta andra konsumenter och producenter och kan
användas för många syften, som att öka tillförlitligheten och/eller öka
elhandeln. Kostnaden för nätförstärkningar är beroende på var vindkraften
lokaliseras relativt elförbrukningen och den existerande nätstrukturen, och
man kan därmed förvänta sig att kostnaden varierar mellan olika länder.
Kostnaden för nätförstärkningar i enligt de studier som sammanställts inom
IEA task 25 visar på en variation från 0 Euro/kW till 270 Euro/kW. Kostnaden
ökar inte kontinuerligt, för vissa enstaka fall kan kostnaden vara mycket hög.
Det är också skillnader mellan hur olika länder väljer att hänföra kostnaden
till vindkraft, eller om transmissionsledningarna även kan användas för andra
ändamål. Det är viktigt att notera att nätförstärkningar ska vägas mot
alternativet att styra vindkraftsproduktionen eller ändra produktionen i andra
kraftslag i de fall nätkapaciteten inte är tillräcklig i vissa situationer. För
transmissionsplanering kan den mest kostnadseffektiva lösningen innebära att
man bygger ut nätet från början för den mängd vindkraft man kommer att se
i framtiden, istället för att bygga ut nätet successivt.
Vindkraftens effektvärde: Vindkraften bidrar också till ett visst
”effektleverensvärde”, dvs elförbrukning kan öka med bibehållen systemtillförlitlighet tack vare vindkraften. Detta värde kan vara upp till 40% av
installerad vindkraftseffekt om vindkraften producerar mycket vid hög
elförbrukning, och ner till 5% vid hög vindkraftsmängd och/eller om lokala
vindar har en negativ korrelation till elförbrukningen. Sammanlagring av
vindkraft i större områden ökar vindkraftens effektvärde. När det gäller
uppskattning av vindkraftens effektvärde används flera olika metoder. Att
uppskatta elsystemets risk för effektbrist (Loss-of-Load-Probability, LOLP) är
den mest noggranna metoden. En viktig faktor är om ägarna till
vindkraftverken erhåller betalning för effektvärdet eller inte. Detta gäller även
andra kraftslag och beror på marknadsstrukturen. Vissa studier använder
termen ”effekt-kostnad”. Definitionen av denna term är kostnaden för att
kompensera skillnaden i effektvärde mellan vindkraft och ett konventionellt
kraftverk. Termen ”effekt-kostnad” används inte så ofta men det är viktigt att
notera att när nivån på denna ska uppskattas ska kompensationen ske i
kraftverk med mycket låg utnyttjningstid, t ex gasturbiner (open cycle gas
turbines, OCGT). Ett alternativ är att kompensera med frivillig lastreduktion.
Båda dessa alternativ har en relativt låg effektkostnad.
ELFORSK
Rekommendationer för integrationsstudier:
(i)
beakta sammanlagringen av vindkraften över större geografiska
områden vilket ger en utjämnad variation av total vindkraft i använda
tidsserier. Förutsätt även att bästa möjliga prognosmetoder för
vindkraftsproduktionen kommer att finnas tillgängliga.
(ii)
undersök vindkraftens variation tillsammans med elförbrukningens
variation. Utgå från historiska data för vindkraften samt elförbrukning
och prognos för denna för att på detta sätt få med en korrekt
beskrivning av samvariationen.
(iii)
beakta elsystemets karakteristik och reaktion på balansförändringar
genom modellering och simulering
(iv)
undersök faktiska kostnader istället för design av tariffer
(v)
jämför kostnader och nytta för vindkraften.
I de flesta fall är frågan om huruvida extra investeringar är lönsamma i ett
elsystem med större mängd vindkraft – inte bara att konstatera att en viss
mängd extra reserver och/eller nya kraftledningar etc är en förutsättning för
att få in mer vindkraft. För integration av stora mängder vindkraft ska
möjligheten till optimering av det integrerade elsystemet studeras. Modifiering
av system-konfigurationen och drift-praxis för att få ett effektivt elsystem kan
vara nödvändigt. Inte alla metoder som används idag är designade för att på
ett korrekt sätt införliva kraftslag såsom vindkraft och metoderna var
ursprungligen heller inte utvecklade för detta ändamål. Genom att öka
systemets flexibilitet genom t ex transmission till angränsande elsystem,
produktionens flexibilitet, elförbrukningens flexibilitet (demand side
management, DSM) och optimalt användande av lager (t ex pumpkraft eller
bränslelager) i kombination med samarbete mellan olika marknader och
driftsplanering närmare faktisk timme påverkas mängden vindkraft som kan
integreras på ett kostnadseffektivt sätt. Det finns en ökad insikt om att
studier av vindkraftens integration bör ske på en internationell nivå för att på
så sätt identifiera behovet och nyttan av internationella kraftförbindelser och
på detta sätt uppnå politiskt satta mål rörande större mängder förnybar
energi.
Framtida arbete: Integration av vindkraft har studerats upp till andelar om
10-20% av årsenergiförbrukningen (upp till 50% av max förbrukning). Vad
som händer vid större andel vindkraft, när vindkraft blir än mer dominerande
del av kraftsystemet, är inte uppenbart – det framtida kraftsystemet kan i
framtiden mycket väl erbjuda olika möjliga alternativ för flexibilitet i
efterfrågan som inte finns idag. Dessutom bör framtida integrationsstudier
beakta en förmodad högre andel av solenergi eller vågenergi och på många
ställen kommer sammanlagring från olika teknologier att minska den totala
variationen. Generalisering av slutsatser för att erhålla grövre uppskattningar
om inverkan av större mängder vindkraft och dess kostnader i olika typer av
kraftsystem kommer att fortsätta vara uppgiften i nästa fas av Task 25.
ELFORSK
ELFORSK
Summary
A R&D Task titled “Design and Operation of Power Systems with Large
Amounts of Wind Power” was formed in 2006 within the “IEA Implementing
Agreement on the Co-operation in the Research, Development and
Deployment of Wind Turbine Systems” (www.ieawind.org) as Task 25. This
R&D task will collect and share information on the experience gained and the
studies made on power system impacts of wind power, and review
methodologies, tools and data used. The results of the first 3-year period is
reported in this report. The work will continue with a second 3-year period.
The following countries and institutes have been involved in the collaboration
(TSO is Transmission System Operator):
•
Denmark: Risø-DTU; TSO Energinet.dk
•
EWEA (European Wind Energy Association)
•
Finland: VTT Technical Research Centre of Finland (Operating Agent)
•
Germany: ISET; TSOs RWE and E.ON Netz
•
Ireland: SEI; UCD; ECAR; TSO Eirgrid
•
Norway: SINTEF; Statkraft
•
Netherlands: ECN; TUDelft
•
Portugal: INETI; TSO REN; INESC-Porto, IST
•
Spain: University Castilla La Mancha
•
Sweden: KTH
•
UK: Centre for Distributed Generation & Sustainable Electrical Energy
•
USA: NREL; UWIG.
The Task started with producing a state-of-the-art report on the knowledge
and results so far, published in VTT Working Papers series in 2007. In the
state-of-the-art report, as well as in this report, a summary of only selected,
recently finished studies is presented. The Task has also started the work of
developing guidelines on the recommended methodologies when estimating
the system impacts and the costs of wind power integration.
In this report only the summary of the results from the IEA Task will be
presented. The full report will be available from
http://www.ieawind.org/AnnexXXV.html
Vindforsk has financed the Swedish participation in this Task. The
participation means that persons from Division of Electric Power Systems,
KTH, Stockholm has participated in meetings, and presented research results
from KTH and other studies performed concerning wind power integration in
Sweden.
ELFORSK
ELFORSK
Innehåll
1 Executive summary from final report.
1 2 Swedish participation.
5 3 Published and presented Swedish reports.
6
Appendix 1. Some Swedish related parts from the final report
ELFORSK
ELFORSK
1
Executive summary from final
report.
Adding wind power will bring about a variable and only partly predictable
source of power generation to a power system that has to balance generation
and varying demand at all times. High penetration of wind power has impacts
that have to be managed through proper wind power plant interconnection,
integration of the generation, transmission planning, and system and market
operations. This final report of Task 25 first term presents a summary of
selected, recently concluded studies of wind integration impacts from
participating countries. The case studies summarized are compared, although
this is not an easy task due to different methodology and data used, as well
as different assumptions on the interconnection capacity available.
There are already several power systems and control areas coping with large
amounts of wind power. Several issues that impact on the amount of wind
power that can be integrated have been identified. Aggregation benefits of
large areas help in reducing the variability and forecast errors of wind power
as well as help in pooling more cost effective balancing resources. An
alternative to large balancing areas is to allow and promote intra-day and
intra-hour trading between different balancing areas in order to obtain lowcost balancing services. System scheduling and operating electricity markets
at less than day-ahead time scales help reduce the forecast errors of wind
power that affect operating reserves. Transmission is the key to aggregation
benefits, electricity markets and larger balancing areas.
For wind penetration levels of 10-20 % of gross demand in power systems,
the cost effectiveness of building new electricity storage is still low (excluding
hydro power with large reservoirs or pumped hydro). With higher wind
penetration levels the extra flexibility that also storages can provide will be
beneficial for the power system operation , provided they are economically
competitive with other forms of flexibility. It is important to notice, however,
that any storage should be operated according to the needs of aggregated
system balancing. It is not cost effective to provide dedicated back-up for
wind power in large power systems where the variability of all loads and
generators are effectively reduced by aggregating, in the same way as it is
not effective to have dedicated storage for outages in a certain thermal power
plant, or having specific plants following the variation of a certain load.
Integration cost of wind power: Many studies address integration costs.
Integration cost is the extra cost of the design and operation of the non-wind
part of the power system when wind power is integrated. Integration cost can
be divided into different components arising from the increase in the
operational balancing cost and grid reinforcement cost. It is important to note
whether a market cost has been estimated or the results refer to technical
costs for the power system. A “market cost” include transfer of money from
one actor to another actor, while “technical costs” implies a cost for the whole
system. Most studies so far have concentrated on the costs of integrating
wind into the power system while also cost-benefit analysis work is emerging.
1
ELFORSK
There is also benefit when adding wind power to power systems: it reduces
the total operating costs and emissions as wind replaces fossil fuels.
Integration costs of wind power need to be compared to something, like the
production costs or market value of wind power, or integration cost of other
production forms. To enable fair comparison between power systems with
differing amounts of wind power, these systems should in principle have same
CO2 emissions, reliability, etc. The value of the capacity credit of wind power
can also be stated.
Increase in short term reserve requirements due to wind power: Wind
generation may require system operators to carry additional operating
reserves. From both the experience and results from studies performed, a
significant challenge is the variability of wind power within 1-6 hrs. Frequency
control (time scale of seconds) and inertial response are not crucial problems
when integrating wind power into large systems at the present time, but can
be a challenge for small systems and will become more of a challenge for
systems with high penetration in the future. The increase in short term
reserve requirement is mostly estimated by statistical methods combining the
variability or forecast errors of wind power to that of load and investigating
the increase in the largest variations seen by the system. The impact of wind
power is mostly seen in the 10 minutes to some hours time scale, and only
little in the second to second automatic frequency control time scale. The
estimated increase in short term reserve requirements in the studies
summarised in this report has a large range: 1-15 % of installed wind power
capacity at 10 % penetration (of gross demand) and 4-18 % of installed wind
power capacity at 20 % penetration. It is of central importance to separate
need of flexibility in longer time scales of several hours to a day (power plants
that can follow net load variation) and need of reserves that can be activated
in seconds or minutes time scale (power plants that can follow unforecasted
net load variations).
An important issue is that “increase in reserve requirements” does not
necessarily mean need of new investments. The amount of wind-caused
reserves is at highest when wind power is on a high production level. In these
situations the other power stations are operated on a low level, which means
that they can act as reserves and increase the generation if wind power
decreases. This must be considered when “increased reserve margins” are to
be estimated. From the cost estimates presented in investigated studies it
follows that at wind penetrations of up to 20% of gross demand (energy),
system operating cost increases arising from wind variability and uncertainty
amounted to about 1–4 €/MWh wind power produced. This is 10% or less of
the wholesale value of the wind energy. The actual impact of adding wind
generation in different balancing areas can vary depending on local factors.
Important factors identified to reduce integration costs are aggregating wind
plant output over large geographical regions, larger balancing areas, and
operating the power system closer to the delivery hour with accurate forecast
systems.
Transmission planning with wind power: With current technology, new
wind power plants are able to meet system operator expectations such as
riding through voltage dips, supplying reactive power to the system,
controlling terminal voltage, and participating in SCADA system operation with
2
ELFORSK
output and ramp rate control. Grid reinforcement may be needed for handling
larger power flows and maintaining a stable voltage, and is commonly needed
if new generation is installed in weak or congested grids far from load centers,
or where no grid exists, such as offshore. Transmission cost is the extra cost
in the transmission system when wind power is integrated. Either all extra
costs are allocated to wind power, or only part of the extra costs are allocated
to wind power –
grid reinforcements and new transmission lines often
benefit also other consumers or producers and can be used for many
purposes, such as increase of reliability and/or increased trading. The cost of
grid reinforcements due to wind power is therefore very dependent on where
the wind power plants are located relative to load and grid infrastructure, and
one must expect numbers to vary from country to country. The grid
reinforcement costs from studies in this report vary from 0 €/kW to 270
€/kW. The costs are not continuous; there can be single very high cost
reinforcements. There are also differences in national studies on how the
costs are allocated to wind power – part of the reinforcements are usually
made also for other reasons than wind power. It is also important to note that
grid reinforcements should be held up against the option of controlling wind
output or altering operation of other generation in cases where grid adequacy
is insufficient during only part of the time or for only some production and
load situations. For transmission planning, the most cost effective solution in
cases that demand considerable grid reinforcements would be to plan and
expand the transmission network for the final amount of wind power in the
system – instead of sequentially planning for multiple phases of incremental
expansion.
Capacity value of wind power: Wind generation will also provide some
additional load carrying capability to meet forecasted increases in system
demand. This contribution can be up to 40 % of installed capacity if wind
power production at times of high load is high, and down to 5 % in higher
penetrations or if local wind characteristics correlate negatively with the
system load profile. Aggregating larger areas benefits the capacity credit of
wind power. Regarding estimating the capacity value of wind power, there are
several approaches used. Determining the Loss-of-Load-Probability (LOLP) of
the power system for different load levels is the most rigorous methodology
available. An important issue is whether wind power owners will be paid for
the capacity value or not. This is also an issue for other types of power plants
and depends on the market regulation. Some reports use the term “capacity
cost”. The definition of this term is the cost for the compensation for the
difference in capacity value for wind power and capacity value for a
conventional power plant. This “capacity cost” is not now in widespread use,
but it is important to note that.when it is calculated this compensation should
be added in power plants with very low utilization time, such as open cycle
gas turbines (OCGT). An alternative is to use voluntary load reduction. Both
these alternatives have comparatively low capacity costs.
Recommendations for wind integration studies: (i) capturing the
smoothed out variability of wind power production time series for the
geographic diversity assumed and utilizing wind forecasting best practice for
the uncertainty of wind power production (ii) examining wind variation in
3
ELFORSK
combination with load variations, coupled with actual historic utility load and
load forecasts (iii) capturing system characteristics and response through
operational simulations and modelling (iv) examining actual costs independent
of tariff design structure and (v) comparing the costs and benefits of wind
power. In most cases the question is whether extra investments to power
systems are economically profitable or not in the new system with larger
amount of wind power – not only stating that a certain amount of extra
reserve capacity and/or new transmission lines etc are a prerequisite in order
to build any wind power.
For high penetration levels of wind power, the optimisation of the integrated
system should be explored. Modifications to system configuration and
operation practices to accommodate high wind penetration may be required.
Not all current system operation techniques are designed to correctly
incorporate the characteristics of wind generation and surely were not
developed with that objective in mind. Increasing power system flexibility
through such means as transmission to neighbouring areas, generation
flexibility, demand side management and optimal use of storage (e.g.
pumping hydro or thermal) in combination with market aggregation and
operation closer to real time will impact the amount of wind that can be
integrated cost effectively. There is growing recognition of the need to assess
wind power integration at the international level to identify the needs and
benefits of interconnection of national power systems in achieving stated
policy goals of accommodating higher levels of renewable energy penetration.
Future work: Wind integration has been studied to wind penetration levels of
10–20 % of gross demand (up to 50 % of peak load). What happens in larger
penetration levels, where wind becomes a more dominating part of power
system, is not completely clear – the future power systems may also provide
different options for flexibility in demand side that do not exist today.
Furthermore, future integration studies should take into account the foreseen
high penetration of PV or ocean power and in similar manner and in many
regions this will help smoothing the variability of individual technologies.
Generalising the findings to give rough estimates for wind integration efforts
and costs for different kind of power systems remains a task for the next
phase of Task 25.
4
ELFORSK
2
Swedish participation.
There have been six meetings in the project:
•
Finland-Helsinki, 24 May, 2006 (Lennart Söder participated). Together
with the Nordic Wind Power conference.
•
USA-Oklahoma, 25 October, 2006 (Lennart Söder participated).
Together with Utility Wind Integration Group (UWIG) meeting
•
Italy-Milano, 7 May, 2007 (Lennart Söder participated). Together with
European Wind Energy Conference, EWEC
•
Norway-Oslo, 19 September, 2007 (Lennart Söder participated),
meetings with Statkraft
•
Denmark-Fredericia, 26-27 March, 2008 (Lennart Söder participated),
meetings with Energinet.dk
•
Ireland-Dublin, 4-5 November, 2008 (Lennart Söder participated)
Lennart Söder has received information concerning activities in other
countries in the meetings, Swedish activities have been presented, and
common issues, such as relevant simulation methods for large scale
introduction of wind power have been discussed.
5
ELFORSK
3
Published and presented Swedish
reports.
Within the area of the Task the following reports have been written, published
and presented in the meetings:
•
Modelling Approach impact on estimation of integration cost of wind
power, Lennart Söder, 7th IAEE, Bergen, Norway, August 2005
•
A comparison of wind integration experiences in some high penetration
areas, Lennart Söder et al, Nordic Wind Power Conference, 22-23 May,
2006, Espoo, Finland
•
Experience From Wind Integration in Some High Penetration Areas,
Lennart Söder, Lutz Hofmann, Antje Orths, Hannele Holttinen, Yih-huei
Wan and Aidan Tuohy IEEE Transactions on energy conversion, vol.
22, no 1, March 2007
•
On methodology for modelling wind power impact on power systems,
Söder, L., Holttinen, H., International Journal of Global Energy Issues,
vol. 29, no 1-2, pp 181-198, 2008
•
A review of different methodologies used for calculation of wind power
capacity credit, Lennart Söder and Mikael Amelin, invited to IEEE
Summer meeting Pittsburg, USA, August 2008
In addition to this, the following Swedish reports have been presented and
discussed in the meetings:
•
Future trading with regulating power, Magnus Brandberg och Niclas
Broman, Uppsala Universitet, Vattenfall Utveckling AB jan 2006 – jun
2006
•
Effects of Large Scale Wind Capacities in Sweden Report 2008-036,
ECON, Report written for the Market Design project:
•
A massive introduction of wind power – Changed market conditions? –
Elforsk report 08:41, Fredrik Carlsson, Viktoria Neimane, Vattenfall,
Sweden, June, 2008
•
Large scale expansion of wind power – Consequences for the
transmission grid and the need of regulating power, Svenska Kraftnät
– 2008-06-01- Dnr 617/2008/AN40
•
The grid connection inquiry – SOU 2008:13, February 2008
•
Nordic grid Master Plan 2008, Nordel, March 2008
•
Coordination of wind- and hydropower in Skellefteå River - Rebnis
and Sädva hydropower plants and the planned Uljabuouda windfarm.
Elforsk report 07:18
6
ELFORSK
•
Research results from KTH projects performed by Elin Lindgren
(Vindforsk), Katherine Elkington (Vindforsk), Julija Matevosyan
(Vindforsk) and Mikael Amelin (Market Design)
•
4000 MW wind power in Sweden, imact of regulation and reserve
requirements, Elforsk report 05:19, Urban Axelsson, Robin Murray,
Viktoria Neimane, Vattenfall Utveckling.
7
ELFORSK
8
Appendix 1.
Some Swedish related parts from the final report
The aim of the IEA Annex XXV is to compare the results concerning wind
power integration studies from different countries. In this chapter the specific
analysis performed concerning Swedish studies is presented. The selected
parts are copied directly from the final report. The full report also includes the
same type of analysis for studies performed in other countries as well as
general descriptions and conclusions. The full report is available from
http://www.ieawind.org/AnnexXXV.html.
Below is extractions from the final report with chapter numbers from the final
report.
3. Balancing and efficiency of production
3.6 Sweden
3.6.1 Reserve requirements, Elforsk 2005
Report: 4000 MW wind in Sweden (Axelsson et al., 2005).
Results: The results are in Table 8. The report neither estimates whether this
increase in reserve requirements could be met with existing capacity, nor
estimates the cost of increased use of reserves. To estimate how much potential
bottleneck situations could affect the results, the same calculation has been made
for different regions in Sweden.
Table 8. Results of increased reserve requirements in Sweden for different wind
power penetrations and different time scale reserves.
Appendix 1. page 1
Input data, wind power modeling: The wind power production input is from a
synthetic time series for years 1996–2001, coinciding with the load data. Load
forecasts were available for 2002–2004 indicating RMSE forecast error of 1.5 %
for short-term forecasts (124 hours) and around 5% for forecasts one week ahead.
Wind power forecasts were assumed reducing the variability to 80 % of
persistence for one hour ahead (from 1.8 % to 1.4 % of installed capacity). For 4
hours ahead, the same level of forecast errors as in Germany were used (2.5 % of
installed capacity). For day-ahead, German data was scaled for Sweden.
Methodology: The methodology for 1 and 4 hour calculations is the same as in
section 3.4.1 (Holttinen, 2004), except that also wind power forecastability has
been taken into account. It can be noted that “stand. dev.” in table 7 means that a
probabilistic method using 4 is applied. For day-ahead, the methodology of
3.7.1 (DENA, 2005) has been used, by scaling the German results to Sweden
assuming similar forecastability of wind power. In the report (Axelsson et al.,
2005) it is stated concerning the use of DENA results that “… .the figures for
Sweden … can probably be considered as an upper limit. It must once again be
noted that also for Sweden these figures relates to the extra requirement during
high wind situations”
3.6.2 Reserve requirements – SvK 2008
Report: Large scale expansion of wind power – Consequences for the
transmission grid and need of regulation power (Svenska kraftnät, 2008)
Results: The amount of needed regulating resources has been estimated based on
the report (Axelsson et al, 2005). The results are shown in Table 9. It can be noted
that the presented amounts are higher than in other reports and the reasons are
explained below.
Appendix 1. page 2
Table 9. Need for additional reserves in Sweden at 4000-12000 MW of wind
power (Svenska Kraftnät, 2008)
Input data, wind power modelling: There are no new data studied in
comparison with the ones used in (Axelsson et al, 2005). However, the results
differ significantly depending on other methods used which interpret the result
differently.
Methodology: The additional primary reserves are calculated in the following
way: From the report (Axelsson et al, 2005) the figure for “Increase in maximum
hourly variation (MW)” is used, i.e., how much the net load (load minus wind
power) changes between two hours will increase. The figure for 6000 MW wind
power is that maximum increase of net load will increase with 199 MW (from
2383 MW to 2582 MW) and maximum decrease of net load will increase with
552 MW (from 1331 MW to 1883 MW). The mean value for positive and
negative “maximum change” is then (199+552)/2= 375.5 MW. At an assumed
production level of 90% of 6000 MW = 5400 MW the “maximum change”
corresponds to 7% of the assumed production level. The “90 % production level”
and “7%” are then used to calculate the upper limit of additional primary reserves
(0.07*4000*0.90=250, 0.07*12000*0.90=750). The lower level is set based on an
assumption of uncertainties in the calculation method.
The additional reserves needed in order to compensate for wind power forecast
errors are based on (Axelsson et al, 2005) where it is stated that in a situation with
4000 MW of wind power the “Day-ahead Max.negative” forecast is 590 MW the
“Day-ahead Max.positive” forecast is 690 MW. The mean value is 640 MW
corresponding to 16% of 4000 MW. At an assumed level of 90% of 4000 MW,
this means a mean day-ahead forecast error of 0.90*0.16*4000 = 576 MW which
in the table is written as 500-600 MW. For a wind power level of 12000 MW, the
Appendix 1. page 3
same method is applied: at 90% of 12000 MW means a mean day-ahead forecast
error becomes 0.90*0.16*12000 = 1728 MW which in the table is written as
1500-1900 MW.
The additional reserves in order to compensate for wind power outages at storm
fronts are estimated as follows: For the upper level a 90% production level in
assumed, and out of this a storm front can affect 25% of the production. This
leads to possible outages of 0.25*0.90*4000 = 900 MW and 0.25*0.90*12000 =
2700 MW respectively. The lower level is set based on an assumption of
uncertainties in the calculation method.
Comments to methodology: The report has used data and results from (Axelsson
et al, 2005) but they have interpreted the results in a different way. Concerning
additional primary reserves for the case 4000 MW of wind power (Svenska
kraftnät, 2008) states 250 MW, while (Axelsson et al, 2005) states 20 MW. The
reason is not different data, but different methods. In the following only the
direction of reserves for “net load increase” will be shown as an example since
“reserves which can increase production” is a generally larger challenge than
“reserves which can decrease production”. (Svenska kraftnät, 2008) uses
“maximum net load change between different hours” as the basic data, even if it is
for estimating the primary reserves, in seconds time scale. It must be noted that
without wind power the maximum load change in an hour is +2383 MW.
And without wind power the requirement for primary reserves in Sweden is 235
MW. With 4000 MW of wind power the “maximum net load change between
different hours” increases with +199 MW, corresponding to an increase of
199/2383 = 8.35%. If the required amount of primary reserves should be based on
“maximum net load change between different hours”, then the required amount of
primary reserves should increase with 8.35%, corresponding to 21 MW, which is
very close to (Axelsson et al, 2005). This means that the method resulting in a
need of 250 MW leads to a significant overestimation of these reserves.
Concerning the additional reserves needed in order to compensate for wind power
forecast errors the extra requirements are based on day-ahead forecast errors. It
must though be noted that the forecast quality improves when one comes closer to
a certain hour. The reserves that have to be kept in a system must be scheduled so
much in advance so their ramp rates can follow the imbalances caused by forecast
errors. In most systems the slowest units can be started up in some hours. In the
Swedish/Nordic system the main part of the reserves is hydro power plants which
can go from stand still to full production in minutes. From the physical point of
view (enough capacity online) it is then not the 24 hour forecast errors that are of
interest but more the 3 hour forecasts for many systems (startup time of 3 hours)
or 1 hour forecasts in the Swedish/Nordic system. The 24hour forecast errors are
interesting from the market point of view (volumes that have to be traded outside
the day-ahead market) but can not be used to estimate required physical MW
Appendix 1. page 4
margins. The use of 24-hour forecast errors leads to an overestimation of required
forecast error reserves.
Concerning the additional reserves in order to compensate for wind power
outages at storm fronts: In the basic data used in (Axelsson et al, 2005), wind
speeds from several places and a certain wind power plant model (cutout wind
speed 25 m/s) have been used. This means that in the resulting wind power
production series, also outages depending on storm fronts are included. In
(Svenska kraftnät, 2008) a 25% outage is assumed. Of course a 25% decrease in
production is possible; the question is though how frequently this happens, the
quality of forecasts and if this means requirements of “additional reserves”. In
(SMHI, 2004), data used in (Axelsson et al, 2005) are presented. In (SMHI, 2004)
it is stated that “Statistics show that a loss of capacity of 50 per cent during a sixhour period happens once a year on average”. This means that loss of 25% of
capacity probably takes 23 hours and happens, probably, 14 times per year.
This means that power plants with enough ramp rates have to be online to
compensate for this power production decrease. (Svenska kraftnät, 2008) do not
consider the possibilities of using forecasts of outages at storm fronts, and it can
be noted that the results mean that more reserves are needed for possible changes
in 23 hours (storm front outages) than what is needed in 24 hours (compensation
for wind power forecast errors). If forecasts are not considered then this leads to
an overestimation of needed reserves to manage outages caused by storm fronts.
Concerning total need of additional reserves the results are obtained using the
method of adding the three types of reserves. It must then be noted that the two
last types of reserves have a very strong interaction since one do not keep these
reserves in different units. It can not be rational to keep margins for possible
changes within 3 hours in some units and at the same time, in other units, keep
margins for what could happen in 24 hours. This means that the method of adding
these three types of reserves gives an overestimation of required reserves.
As stated in section 3.2: In a situation with high wind power production, i.e., a
situation with possible decrease of wind power production, other units are not in
operation (since some of the load is met with wind power). This means that these
units can increase their production if wind power decreases and/or load increases.
The consequence is then that it is very common that needed reserves for wind
power decrease are often available in the same way as power plants are available
in a low load situation to meet load increase. In (Svenska kraftnät, 2008) there is
no study concerning the coupling between needed reserves and the availability of
these. This will require much more detailed studies of the future system.
In the report the total costs of reserves is estimated as the calculated total amount
of reserves multiplied with the costs per MW that the TSO currently pay for the
peak load capacity (capacity costs for production units and flexible loads used,
perhaps, some single hours per year). This results in an overestimation of the cost
Appendix 1. page 5
since the total amount of reserves is overestimated. If these reserve units need any
kind of “subsidy” (as peak load capacity) depends on how the prices are set on the
market. There are currently no capacity payments at all to operating reserves in
Sweden.
3.6.3 Imbalance costs for wind power producers
Report: “A massive introduction of wind power. Changed market conditions?”
(Neimane & Carlsson, 2008)
Results: In this report eight different actors have been created, that all have
balance responsibility for their production, which means that if they cause an
imbalance they have to pay up or down regulating prices. These actors are
different in the sense that some are small and some are big, some have
concentrated wind farms and some have wind farms that are geographically
spread-out. In the study the imbalance costs for these actors have been estimated
for a future scenario with larger amounts of wind power. The result is shown in
Fig 25, where Elbas is the intra-day market closing one hour before delivery, for
each hour. It is assumed that wind power will increase with 4000 MW compared
to 2006. This implies that the total imbalance will increase from 0.95 TWh/year to
1.7 TWh/year.
Fig 25. Comparison on the cost of forecast errors on a) the 2006 market, b) the
future market and c) future market and acting on Elbas. 1 SEK = 0.1 Euro.
Input data, wind power modelling: Forecast error data from the 160 MW wind
power farm Horns Rev for the period 11 September 2006 to 31 March 2007 has
been made available for this study, but the data is used taking limited forecast
error correlations into consideration. Regulation prices for the year 2006 are used
Appendix 1. page 6
Methodology: To calculate the cost of the forecast errors, a developed price
model by Klaus Skytte at Risø Laboratory in Denmark has been used. This model
has parameters that have been estimated for the market situation during 2006. By
generating the forecast errors for all actors as random numbers with normal
distribution in Excel for a whole year, it has been possible to calculate the actors’
cost for their forecast errors. It can be noted that imbalance costs are paid between
different market actors, and one can not draw the conclusion that an increased
imbalance cost automatically corresponds to an increase of total cost.
3.6.4 Increase in the use of reserves
Report: Future trading with regulating power, (Brandberg & Broman, 2007)
Results: The purpose has been to investigate how the Nordic regulating
power market will react to integration of 4000 MW of wind power in Sweden.
Results from this study by using two different methods are presented in the
table below.
Input data: Data from 2003 for the West Danish power system with an installed
wind power capacity of 2400 MW have been used. The data consists of 24-hour
and 4-hour forecasts and actual production of wind power
Methodology: The method used for investigating the impact on the regulating
market prices is by studying the impact of wind power forecast errors. The
forecast errors for wind power production have been added to historical regulating
quantities and the new prices have been estimated according to the new regulating
power quantity levels. Wind power production forecast errors have been estimated
in two different ways:
1. The forecasts errors have been calculated and scaled up to reflect an
installed capacity of 4000 MW of wind power. The 24-hour forecast errors
have been used to estimate the increase of adjustment power on the Elbas
market, and the 4-hour forecast for increase of regulating power.
2. Calculation of forecast errors by setting the forecast errors to the change in
production between the hour prior to the hour of operation and the hour of
operation. This forecast error has also been calculated for the Danish data
and scaled up to 4000 MW.
Appendix 1. page 7
3.6.5 Efficiency of hydro power
Report: Integration study of small amounts of wind power in the power system
(Söder, 1994).
Results: Swedish wind power installations of about 2–2.5 TWh/year do not affect
the efficiency of the Swedish hydro system. At wind power levels of about 4–5
TWh/year the installed amount of wind power has to be increased by about 1 % to
compensate for the decreased efficiency in the hydro system. At wind power
levels of about 6.5–7.5 TWh/year the needed compensation is probably about 1.2
%, but this figure has to be verified with more extended simulations.
Input data, wind power modeling: Many generated power series based on
stochastically generated wind speed forecast errors.
Methodology: Wind power balancing was performed in one river using a detailed
model including station efficiencies and the result was upscaled to Sweden.
Deterministic planning but evaluation based on stochastic forecast errors. The
”integration cost” was calculated as needed extra energy (MWh) to compensate
for lost hydro energy. The weekly load was increased in order to compensate for
mean wind energy increase. Load and wind uncertainty were treated. Wind power
was increased until evaluation strategy did not work. Review Table is in
Appendix 2.
Assumptions: Best available wind speed forecasts (in 1994) assumed available.
Rescheduling of hydro plants assumed every hour to consider new improved wind
speed forecasts. Full access to a grid assumed, i.e., no limitations and 100%
reliable.
Limitations: All Swedish wind power assumed to be balanced only with Swedish
hydro power. Trading with neighboring systems and thermal power operation
assumed unchanged. Results origin from study of a smaller part of the Swedish
system and scaled up to be representative. Changes in electrical grid losses not
considered.
Appendix 1. page 8
4. Grid reinforcement and efficiency
4.9 Sweden
PhD study by Julija Matevosyan “Wind power integration in power system with
transmission bottlenecks”, 2006. Study: Economical evaluation of the value of
transmission expansion to limit wind power spillage. (Matevosyan, J. 2006)
Results: With no grid extension the spillage resulting of different wind power
installations in northern Sweden are presented in Table 16.
Table 16. Estimated wind energy spillage in North Sweden without grid
investments, calculated with two methods.
Fig 35. The cost for the spillage as a function of installed amount of wind power
for two price levels of the spillage.
4000 MW of wind power will lead to 15.4% of wind energy being curtailed with
cost of approximately 540 MSEK/year with the curtailment cost of 0.4 SEK/kWh.
For 3200 MW it is 300 MSEK/year. Consequently, a new 800 MW transmission
line decreases costs for energy spillage to 540–300 MSEK/year. The cost for the
needed 800 MW transmission line is for this case 400 MSEK/year. For this case it
is therefore not motivated to build a new line just to motivate lower wind energy
spillage. It can though be noted that a required interest rate of 10% has been used
for this transmission line investment which is comparatively high.
Appendix 1. page 9
Input data, modelling – Line cost data, current use of actual transmission
corridor, interest rate for transmission line investment, duration curve of possible
wind power production. Cost of wind energy spillage.
Methodology – Simplified calculations based on duration curves and available
time series.
Assumptions – No use of local hydro power storage (or any other kind of local
use of wind power) in Northern Sweden to store excess wind power. If this is
considered (which is done in the thesis) then the value of increased transmission
will decrease even more.
Limitations – A specific study for this certain region, where an extension of a
large corridor with 7000 MW capacity is considered.
6. Experience from operating power systems
with large amounts of wind power
6.5 Sweden: Gotland
•
All balancing in the island is done with the HVDC cable to the mainland.
When wind power penetration (of gross demand) exceeded 10 % there
were occasional curtailments when wind power production was close to
the local load and the cable was run near 0 MW. To overcome this
problem, control of the cable was enhanced to enable switching to export
and import near 0 MW. After this there has not been any need to curtail
wind power. From mainland side point of view Gotland wind power
production is comparatively small (Söder et al., 2007).
Appendix 1. page 10
Download